SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Mistral is hiring an LLM Scientist to advance AI-driven scientific discovery. You will develop and optimize large language models, agentic AI systems, and deep learning frameworks to solve open problems in natural sciences including biology, chemistry, physics, and materials science.
Key responsibilities include:
• Participate in end-to-end training of LLM-based solutions with focus on scalability, generalization, and scientific applicability
• Design original workflows to accumulate high-quality scientific data at scale through synthetic or crowdsourced pipelines
• Develop and improve reinforcement learning and agentic systems tailored to natural sciences challenges (e.g., modeling resolution, material property prediction, advanced data analysis)
• Work closely with domain scientists to identify AI opportunities and refine models based on feedback
• Contribute to company infrastructure to improve usability for model training in scientific contexts and develop open-source AI tools for science
• Set research agenda for AI in science, exploring high-risk/high-reward ideas
Mistral provides full-stack AI solutions from frontier models to developer tools and applications. The company partners with enterprises in finance, manufacturing, defense, healthcare, and the public sector to co-create customized AI systems. Teams are distributed across Europe, North America, Asia, and the Middle East.
Requirements:
• Strong background in LLMs, deep learning, or agentic AI with track record of real-world applications (publications, open-source contributions, or deployed systems)
• Extensive experience in computer science; ideally experience scaling synthetic data generation
• Familiarity with latest advancements in LLM architectures and post-training techniques
• Hands-on experience with PyTorch, TensorFlow, or JAX; proficiency in model optimization and scaling techniques (fine-tuning, RLHF)
• Ability to understand at least one natural science domain at high level (graduate-level specialization plus general undergraduate knowledge)
• Proficiency in scientific computing (numerical methods, HPC) and software best practices
• Experience translating technical AI concepts for non-experts and iterating based on user needs
• Team player with low ego and willingness to tackle operational challenges (data pipelines, model deployment)
